Genetic Algorithm for Arabic Word Sense Disambiguation

Word sense disambiguation (WSD) problem asks to assign a meaning to a word according to a context in which it occurs. Many solutions exist for WSD in natural languages, such as English, but research work on Arabic WSD (AWSD) remains limited. AWSD is a more exigent task because Arabic has an intrinsic complexity in its writing structure and ambiguity, such as syntactic, semantic, and anaphoric ambiguity levels. Genetic algorithms (GAs) can be effective to solve this problem since they have been successfully used for many NP-hard optimization problems. In this paper, we propose a new approach to solve AWSD problem based on a GA. We describe a prototype of AWSD system in which we test the performance of our algorithm by carrying out experiments on Arabic sample text, and compare it with a naïve Bayes classifier for AWSD. We show the benefit of the proposed approach and its advantage over naïve Bayes classifier.